Efficient Memory Management for Large Language Model Serving with PagedAttention
1.4 هزار بار بازدید -
10 ماه پیش
-
The paper proposes PagedAttention, an
The paper proposes PagedAttention, an attention algorithm inspired by virtual memory and paging techniques, to address the memory inefficiencies in large language model serving systems. The proposed system, vLLM, achieves near-zero waste in memory and improves throughput by 2-4 times compared to existing systems.
https://arxiv.org/abs//2309.06180
YouTube: @arxivpapers
PODCASTS:
Apple Podcasts: https://podcasts.apple.com/us/podcast...
Spotify: https://podcasters.spotify.com/pod/sh...
https://arxiv.org/abs//2309.06180
YouTube: @arxivpapers
PODCASTS:
Apple Podcasts: https://podcasts.apple.com/us/podcast...
Spotify: https://podcasters.spotify.com/pod/sh...
10 ماه پیش
در تاریخ 1402/06/21 منتشر شده
است.
1,463
بـار بازدید شده